Overview
The work is divided into 5 tasks with subtasks that are described in the following. Most of the experimental equipment and methods are already at hand. The largest uncertainty for the progress of the project is how well the CFD models will predict the experiments. It is the intention to find computational models that are physically founded and thus apply to a wider range of combustion conditions. But as a backup plan it may be necessary to move toward more empirical models that rely more on calibrations.
Task 1: Computational modeling of radiation in IC engines using CFD
OpenFoam will be used as CFD software during the project. It is open software that offers full access to modify the program code. It is free of charge and can be used unlimited for parallel computation on computer clusters. This is essential for development of a detailed engine model. The model development will take place at DTU MEK and will be carried out by a Senior Scientist and a Post Doc. Modification of the program code will be made in collaboration with POLIMI. The combustion and soot models will be implemented in collaboration with NTNU and the detailed engine model together with MDT.
Task 1.1 Gaseous fuel flame models:
Computational models for test cases from Task 3 with gaseous fuel flames will be developed. Combustion and soot models will be tested and validated against measurements from Task 3. Computational interface for inclusion of data from the lumped spectra databases developed in Task 2.2 into radiation models integrated in OpenFOAM will be developed. The result will be a set of model cases that are ready to test modifications of the radiation model during the project.
Task 1.2 Liquid fuel flame models:
Computational model of spray bomb experiments in Task 4.1 will be developed. Numerical predictions of changes in spray parameters, like bulk density change from atmospheric air to engine conditions, will be directly compared to the experiments. Data from the atmospheric spray chamber experiments will be used to fine tune the computational spray model and enable it for calculations on full size fuel injector of MDT test engine.
Computational models for experimental test cases from Task 4.2 with liquid fuel flames will be developed. A special focus will be on finding the most adequate approach for coupling the radiant heat transfer model with the fuel spray model. Spectral data of liquid and vaporized fuels from Task 2.3 will be used.
Task 1.3 Detailed diesel engine model:
Computational model of the full scale MDT test engine will be developed. It will be based on existing engine models but further extended with the developed radiation model. Also the combustion, soot and spray models from Task 1.2 and 1.3 will be tested. Available experimental data will be fully integrated into computations through careful selection of boundary conditions (BC) applied. This is important to ensure that any uncertainty about applied BC’s is removed from the obtained results, thereby enabling a direct quality assessment of the developed radiation model. The BC at the scavenge air inlet will be based on the results of another active project between DTU MEK and MTD where basic properties of swirling scavenge air are investigated. The BC at the fuel injector will be taken from the work in Task 1.2. The BC’s at the cylinder walls will be based on the thin film thermocouple measurements from Task 5.2.
Task 2: Development of the spectral model
A detailed spectral model for gas radiation has been developed at RISØ DTU. The model will be validated against measurements in a unique, ceramic, high temperature gas cell facility at RISØ DTU. Benchmarking with other spectral models like HITEMP will be performed as well.
Task 2.1 Database of high temperature gas spectra:
A database of detailed CO, CO2 and H2O spectra for different temperatures and pressures will be developed from the spectral model. It will be validated experimentally up to 1600 °C which is the upper limit for the gas cell. The investigations should justify using of the database up to 2500 °C and 200 bar (engine conditions).
Task 2.2 Procedure for lumping spectra:
A method for lumping the detailed gas radiation data along with continuous soot radiation should be developed. A simple 1D discrete radiation transfer model will be used to validate the lumping procedure against line of sight measurements from Task 3.
Task 2.3 Database of liquid and gaseous fuel spectra:
A database with spectra of relevant liquid and vaporized fuels will be developed experimentally. To the extent that is experimentally possible, temperatures should range from atmospheric temperature to the boiling point of the liquids and to the cracking temperature of the gases.
Task 3: Measurement of gaseous fuel flames
A carefully chosen selection of laboratory flames suitable for model validation will be mapped. The measurements will be carried out at DTU MEK where most of the equipment and expertise is already present. Sandia will contribute with expertise in flame control and measurement techniques. The flames will be based on well defined hydrocarbon gaseous fuels that may produce sooting as well as non sooting flames, but always will produce CO2 and H2O radiation. The selection of flames will represent an incremental increase in complexity that comprises: laminar premixed flat flames, turbulent premixed flat flames, laminar diffusion flames, turbulent diffusion flames. This incremental approach is essential to identify possible deficiencies in CFD modeling.
Task 3.1 Laminar premixed flat flame experiments:
This laboratory flame offers the highest level of flame control. The fuel to oxygen ratio, inert gas content, flame speed and initial gas temperature is directly controlled by the gas mixer and the burner cooler. It also offers the best access for flame measurements both optically and by probing. With co-flow of inert gas soot oxidation is omitted. Effects of diffusion are minimized due to the premixing. The sum of all these features together with the omitted turbulence makes the flames ideal for validating the reaction kinetics of the combustion and soot formation models. Furthermore, the measured temperature profiles may be used to force the temperatures in the computational model to correct values. This will diminish the influence of heat transfer model while only the effect of the combustion and soot formation model remains.
Task 3.2 Turbulent premixed flat flame experiments:
This laboratory flame offers the same features as in Task 3.1 except that turbulence is now an adjustable parameter. The flame speed is highly sensitive to turbulence because the upstream heat transfer approximately scales with the turbulent viscosity. Adding turbulence to the flame allow for validation of the turbulent flame speed that is predicted by the model.
Task 3.3 Laminar diffusion flame experiments:
The laminar diffusion flame is less dependent on the combustion model while diffusion controls most of the flame properties. For laminar flames only molecular diffusion exists which is much easier to deal with than turbulent diffusion. Assuming that the molecular diffusion is correctly predicted will leave the bulk gas velocity on the fuel rich side of the flame to control the location and shape of the flame envelope. The cold gas velocity is controlled by the burner setup thus acceleration is caused by heat transfer from the flame envelope to the interior gasses. Further assuming that heat conduction and buoyancy is correctly predicted isolates the radiant heat transfer to the flame interior as the process that controls the size and shape of the flame envelope. This feature causes the laminar diffusion flame to be important for validating the radiation model.
Task 3.4 Turbulent diffusion flame experiments:
This laboratory flame offers many of the same features as in Task 3.3 except that turbulence is now an adjustable parameter. The location of the flame envelope is highly sensitive to turbulence because the diffusion approximately scales with the turbulent viscosity. Adding turbulence to the flame allow for validation of the turbulent diffusion that is very important in diesel engine flames. Furthermore the experiment is important for validation of the NOx formation and soot oxidation.
Task 4: Measurement of liquid fuel flames
Task 4.1 Spray bomb experiments:
Fuel spray measurements will be performed in a high pressure bomb at either SANDIA or DTU MEK. The bomb enables spray measurements in cold or hot bulk gas at high pressures either with or without combustion. The fuel injector will be smaller than the MDT test engine injector. But measurements at relevant temperatures and pressures will enable to validate the spray models ability to extrapolate the full scale spray measurements in Task 5.1 from atmospheric to engine conditions.
Task 4.2 Liquid fuel flame experiments:
An experimental setup with a turbulent diffusion flame for well defined liquid hydrocarbon fuels will be developed. It should comprise the features of a true diesel engine flame as much as possible. But it should be stationary and have a simple geometry to enable accurate measurements and faster CFD computations. Measurements of the flame will provide data for validating the radiant heat transfer between the flame and the fuel spray. This transfer is very important because it causes a feedback by increasing the evaporation rate of the fuel that again increases the flame radiance and radiant heat transfer back to the spray. The radiant feedback differs from the convective/conductive feedback by being much less sensitive to the distance between flame and droplets, thus allowing the flame envelope to grow larger.
Therefore the radiant feedback significantly promotes the heat release rate and the size of the flame envelope.
Task 5: Measurements on the MDT test engine
Measured data of the MDT test engine is needed for development and validation of the detailed engine model. MDT has a large test engine database already but some data relevant to this project is still missing. Methods and equipment have been developed in earlier projects by MDT, DTU-MEK and RISØ DTU but a significant effort needs to be made in order to combine them to produce relevant data needed for this project.
Task 5.1 Time resolved flame radiance spectra:
In-cylinder measurements will be performed with a spectrometric technique that covers a wide spectral range of light emission from diesel engine flames. This comprises soot radiation (VIS and near-IR) as well as gas radiation (mid-IR). The method is fast enough to resolve the transient radiation from a slow running marine diesel engine. The field of view is narrow, so that the spectrum is collected from a well defined line of sight. The measurements will provide time resolved spectra of flame radiance that will be used to validate the CFD model directly. Furthermore, the data may be used to determine the time resolved flame temperature and radiant heat transfer to the wall that will be used for model validation as well. As a supplement tests may be performed at MDT fuel injection testrigs to measure specifically the absorptions characteristics of large engine fuel sprays.
Task 5.2 Time resolved wall temperature and heat flux:
Thin film thermocouples with thermal conductivity and absorbance that is equivalent with the wall material will be used at relevant locations in the combustion chamber. The acquired temperatures provide important information for setting accurate, time resolved wall boundary conditions in the CFD model. The measured heat flux comprises both convective and radiant heat transfer from the gases. By subtracting the radiant heat transfer measured in Task 5.3 a measure of the time resolved convective heat transfer is also provided. This is important for validating the convective heat transfer near the wall which is always critical with turbulence models.